Development of the predictive maintenance methods with using computer vision and internet of things

With the increasing use of high-speed trains in railways, there is a greater need for safety and maintenance planning in this area. For this reason, interest in terms such as condition monitoring, fault diagnosis and predictive maintenance has also increased. In recent years, with the development of technology, artificial intelligence, image processing and smart computing techniques are mostly preferred in railways in defect and wear detection, condition monitoring and maintenance planning, as in every field. In this thesis, it is aimed to determine the faults and wear occurring in the pantograph that supplies power to the locomotives from the catenary line in electric trains and to determine the predictive maintenance methods. For this reason, detailed information has been obtained by scanning publications related to these issues in the literature on pantograph and catenary line. Methods created by using image processing and artificial intelligence techniques have been developed to detect faults and wear on the pantograph. The data used in this thesis studies conducted in the Republic of Turkey State Railways (TCDD) 5th Regional Directorate and Railway Research and Testing Center (DATEM) has been obtained in the field work done. In addition, it has been used in data accessible on the internet. Using these data, fault diagnosis and predictive maintenance methods have been performed using methods such as image processing, computer vision, deep learning and augmented reality. The results obtained from the developed methods are compared with the studies in the literature, and the results are given in detail and successful and effective results have been obtained.